Participants and the web-based survey
Study participants were recruited between September 2017 and January 2018 from university and vocational school student cohorts majoring in nutrition, medical staff, local area residents, and from personal networks from several regions of Japan. A total of 398 individuals responded to a web-based survey. Nineteen respondents were excluded as they did not meet the lower age limit of at least 20 years. The final number of adults enrolled in this study was 379.
The web-based survey was a self-administered questionnaire that assessed demographic parameters (age, body mass index, marital status, employment status, household income, residential situation), vegetable intake per meal (breakfast, lunch, dinner, snack), stage of change (precontemplation, contemplation, preparation, action/maintenance), decisional balance (perceived benefits and barriers), the importance of vegetable intake, and confidence (self-efficacy) in eating five SVs of vegetables per day.
The study protocol conformed to the ethical guidelines of the Helsinki Declaration, and the study design was approved by the ethics committee of Kyoto Medical Center (18–095). Informed consent was obtained from all participants prior to starting the survey.
Factor Analysis
The questionnaire was evaluated by five dietitians and three researchers to assess clarity and was modified on the basis of their comments.
After careful consideration of the input from the experts, and an examination of the existing scales, an exploratory factor analysis was performed. Its purpose was to assess the validity of the items in the questionnaire and to identify the factors that group the items. The questionnaire initially included 15 items related to perceived benefits and 15 items related to perceived barriers. Each item was scored on a 5-point Likert-like scale ranging from “strongly disagree” (1 point) to “strongly agree” (5 points).
The factor analysis involved Promax rotation of the retained items. Promax rotation maximizes the sum of the variance of the squared loadings, where loading is defined as the correlation between the item and the factor. It highlights a small number of key items, which simplifies the interpretation of the results. Interpretation of the rotated value involves the identification of the items loaded on each retained factor, the determination of the conceptual meaning of the items loaded on the same factor, and the conceptual differences between the items loaded on different factors. The relationship of each item to the underlying factor is expressed by the rotated factor loading value. Pattern loadings with an absolute value of approximately 0.35 or more were used to interpret the results.
Internal Validity
The items assessing perceived benefits and perceived barriers were tested for construct validity using the Kaiser–Meyer–Olkin (KMO) test. A KMO value of approximately 1 is good, and a value of 0.6 is acceptable. Items with KMO values < 0.5 (which indicates low communality) can usually be excluded from the analysis [13]. The reliability of the questionnaire was evaluated using Cronbach’s alpha coefficient [14], which scores internal consistency as follows: <0.5, unacceptable; 0.5 to < 0.6, poor; 0.6 to < 0.7, questionable; 0.7 to < 0.8, acceptable; 0.8 to < 0.9, good; and ≥ 0.9, excellent.
External Validity
Regarding the stage of change for vegetable intake, 10.6% of the participants were in precontemplation, 63.6% were in contemplation, 8.7% in preparation, and 17.1% in action/maintenance. The numbers of SVs for the stages were 2.04, 2.23, 2.67, and 4.22, respectively (P for trend < 0.001). For each meal except snacks, vegetable intake increased across the stages (P for trend < 0.001) (Table 2).
Figure 2. Pros and cons T-scores by stage of change.
The scores were presented as standardized T-scores (mean = 50, SD = 10).
Table 2
Vegetable Intake and Decisional Balance Score by Stage of Changes
Stage of changes | All (n = 379) | PC (n = 40) | C (n = 241) | P (n = 33) | A/M (n = 65) | p |
Vegetable intake, SV | | | | | | |
Breakfast | 0.36 (0.48) | 0.19 (0.29) | 0.26 (0.38) | 0.41 (0.46) | 0.78 (0.63) | < 0.001 |
Lunch | 0.84 (0.58) | 0.81 (0.55) | 0.73 (0.53) | 0.89 (0.46) | 1.24 (0.64) | < 0.001 |
Dinner | 1.35 (0.79) | 1.04 (0.58) | 1.20 (0.68) | 1.33 (0.55) | 2.11 (0.95) | < 0.001 |
Snack | 0.05 (0.19) | 0.00 (0.00) | 0.04 (0.18) | 0.03 (0.12) | 0.10 (0.28) | 0.013 |
Total / Day | 2.59 (1.46) | 2.04 (1.12) | 2.23 (1.21) | 2.67 (0.93) | 4.22 (1.55) | < 0.001 |
Benefits (pros), points Total Pleasure Healthy | 4.00 (0.60) 3.92 (0.71) 4.08 (0.61) | 3.66 (0.62) 3.55 (0.74) 3.77 (0.64) | 3.95 (0.59) 3.85 (0.71) 4.04 (0.62) | 4.16 (0.42) 4.12 (0.48) 4.21 (0.48) | 4.32 (0.52) 4.31 (0.58) 4.33 (0.52) | < 0.001 < 0.001 < 0.001 |
Barriers (cons), points Total score Not attractive Low priority Hard to get | 2.57 (0.76) 2.81 (0.82) 2.16 (0.96) 2.30 (1.09) | 2.94 (0.61) 3.16 (0.64) 2.56 (0.93) 2.75 (1.19) | 2.67 (0.72) 2.92 (0.77) 2.26 (0.97) 2.40 (1.09) | 2.43 (0.71) 2.69 (0.82) 2.10 (0.88) 2.02 (0.91) | 2.02 (0.77) 2.27 (0.88) 1.58 (0.76) 1.79 (0.94) | < 0.001 < 0.001 < 0.001 < 0.001 |
Values are the mean (S.D.). The decisional balance score consisted of twelve pro and twelve con items scored from 1 to 5. PC (precontemplation, n = 40), C (contemplation, n = 241), P (preparation, n = 33), A/M (action/maintenance, n = 65).
At each stage, there was a significant difference between the T-scores for the pro items and con items (Fig. 2).
The decisional balance score (the pro score minus the con score) was 0.72 for precontemplation, 1.28 for contemplation, 1.73 for preparation, and 2.30 for action/maintenance (P for trend < 0.001). It increased in both pro domains and decreased in all three con domains across the stages (P for trend < 0.001) and positively correlated with vegetable consumption (Spearman's correlation: 0.461; P < 0.001). The number of vegetable SVs was directly proportional to the perceived benefits score and inversely proportional to the perceived barriers score (Spearman's correlation: 0.324 and 0.435, respectively). Higher vegetable consumption correlated with lower perceived barriers scores (odds ratio: 0.42; 95% CI: 0.26, 0.68) and higher perceived benefits scores (odds ratio: 2.75; 95% CI: 1.40, 5.40).
For each increase in the number of vegetable SVs, the pros score increased by 0.76 (standard error: 0.12), whereas the cons score decreased by 0.77 (standard error: 0.09). Interestingly, the effect sizes for the pro and con items were similar (Hedges' g: 1.18 and 1.29, respectively).
Reliability Testing
Intraclass correlation coefficients (ICCs) with 95% confidence intervals (CIs) were used to assess test-retest reliability by comparing the participants’ responses to the questions at two time points. The interval between the test and retest was at least one week but less than four weeks. The ICCs were calculated using a 2-way mixed model based on absolute agreement. Agreement was rated as suggested by Landis and Koch [18] as follows: <0.00, poor; 0.00–0.20, slight; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, substantial; and 0.81–1.00, almost perfect.
Statistical Analysis
Each of the final 12 items in the pro and con categories was summed and standardized using T-scores for ease of reporting. A trend analysis was used to examine the relationship of decisional balance to the stage of change. Cronbach’s alpha coefficient was used to determine the reliability of the factors (internal consistency). ICCs were used to evaluate reproducibility. P values < 0.05 were considered significant. Hedges' g was used to categorize effect sizes as small, medium, or large (g = 0.20, 0.50, and 0.80, respectively). All statistical analyses were performed using SPSS for Windows (IBM SPSS Statistics, ver. 20.0).